If the phone itself might be compromised, does end-to-end encryption in a messaging app buy me anything?

What survives, even assuming the endpoint is untrustworthy:

• Protection against everyone else. The network operator, the service provider, anyone on the wifi, anyone who obtains the server's storage. That is a large set of adversaries and they are eliminated.
• Protection for the other participants' other conversations. A compromise of your device does not expose their messages with anybody else.
• A limit on retroactive exposure. Modern protocols rotate keys as the conversation proceeds, so obtaining a key at one moment does not open the whole history.

So the correct framing is not "encryption is useless if the device is compromised". It is "encryption changes your problem from many adversaries to one". That is a genuine improvement, and it is also a warning that the one remaining adversary is now the whole game.

The practical mistake people make is assuming the app is where the security lives, and therefore that the device does not matter.

For the last part of your question — the honest answer for someone who genuinely needs this — the work moves entirely to the device.

• Use a device you have reason to trust, from a vendor with a verifiable update record and a security model you can read about. Trust in hardware is not binary but it is not uniform either.
• Keep it updated. The overwhelming majority of real device compromises use vulnerabilities that were patched.
• Reduce what is installed. Every app is a potential problem, and side-loaded software from unknown sources is the single largest risk factor for ordinary users.
• Use a separate device for the sensitive communication if the stakes justify it, doing nothing else on it.
• Verify contacts in person using the app's safety number or equivalent, which protects against a different attack — someone substituting themselves for your contact.

And be realistic about the threat. If your adversary is a well-resourced actor specifically targeting you, no consumer advice in this thread is sufficient and you need help rather than a checklist.

The short version, and it is worth memorising because it settles a lot of questions in this area: anything you can read, a compromised device can read.

Encryption protects the message in transit and at rest. It cannot protect the moment where the message is plaintext, and that moment necessarily exists at both ends — on the screen you read it from and on the keyboard you type it into. Software with sufficient privilege on the device sits inside that boundary, not outside it.

So against a genuinely compromised endpoint, end-to-end encryption provides no protection for the content of your conversation. This is not a weakness in the design; it is a statement about where the design's boundary is, and the boundary was never claimed to include the device.

What it still gives you, even on a suspect device, is worth listing rather than dismissing.

One nuance on "untrusted manufacturer": there is a meaningful difference between hardware you distrust and software you distrust.

A backdoor in silicon is extraordinarily difficult to build, difficult to keep hidden, and rarely necessary. The overwhelming majority of real device compromise is at the software layer — the operating system image, pre-installed applications, an update channel.

That matters because software problems are the ones you can do something about, by choosing what you install and whether the device receives updates. Worrying about the chip while installing arbitrary applications is the wrong order of concern.